Montpellier, France
Current and projected computing technologies, e.g. machine vision, massive storage, parallel processing, neural nets, fuzzy logic, etc., can be combined into sophisticated and intuitive software for ethology based applications. Developing flexible self-improving systems which automatically adapt to environmental/experimental constraints is possible.
Many questions must be asked, problems solved and mistakes made before dynamic, versatile and robust machine vision and software systems become indispensable aids for ethologists. Questions concerning what the programs will perform, which methods they will employ and how the programs can be validated remain to be explored.
The author suggests a global approach in order to integrate independent technical compo-nents and scientific know-how. This paper provides suggestions and approaches for developing automated behavioral analysis tools using evolving computing technologies.
During the workshop, the program SUIVI will be demonstrated. SUIVI was designed to explain and illustrate the problems associated with converting animal motion to data and to demonstrate the author's approach of dividing the problem into input, data and behavior analysis.